electron-phonon coupling, thermal expansion coefficient
TRANSCRIPT
1
Electron-Phonon Coupling, Thermal Expansion Coefficient, Resonance Effect
and Phonon Dynamics in High Quality CVD Grown Mono and Bilayer MoSe2
Deepu Kumar1#, Vivek Kumar1, Rahul Kumar2, Mahesh Kumar2, Pradeep Kumar1*
1School of Basic Sciences, Indian Institute of Technology Mandi, 175005, India
2Department of Electrical Engineering, Indian Institute of Technology Jodhpur, 342001, India
Abstract
Probing phonons, quasi-particle excitations and their coupling has enriched our understanding of
these 2D materials and proved to be crucial for developing their potential applications. Here, we
report comprehensive temperature, 4-330 K, and polarization-dependent Raman measurements on
mono and bilayer MoSe2. Phonon’s modes up to fourth-order are observed including forbidden
Raman and IR modes, understood considering Fröhlich mechanism of exciton-phonon coupling.
Most notably, anomalous variations in the phonon linewidths with temperature pointed at the
significant role of electron-phonon coupling in these systems, especially for the out-of-plane
(1gA ) and shear mode ( 2
2gE ), which is found to be more prominent in the narrow-gaped bilayer than
the large gapped monolayer. Via polarization-dependent measurements, we deciphered the
ambiguity in symmetry assignments, especially to the peaks around ~ 170 cm-1 and ~ 350 cm-1.
Temperature-dependent thermal expansion coefficient, an important parameter for the device
performance, is carefully extracted for both mono and bilayer by monitoring the temperature-
dependence of the real-part of the phonon self-energy parameter. Our temperature-dependent in-
depth Raman studies provide a pave for uncovering the deeper role of phonons in these 2D layered
materials from a fundamental as well as application point of view.
#E-mail: [email protected]
*E-mail: [email protected]
2
1. Introduction
Group VI transition metal dichalcogenides (TMDCs) in their naturally occurring bulk form have
been studied for decades due to their rich physics and industrial applications [1-2]. Research
interests in these TMDCs materials have gained a lot of attention after the successful isolation of
single-layer graphene from graphite in 2004 [3]. These are the layered materials belonging to the
family of 2D materials with a common atomic formula MX2, where M is the transition-metal atom
(Mo or W), and X is the chalcogen atom (S, Se or Te) [4-6]. The properties of these 2D materials
are strongly dependent on the number of layers[7-8]. For example, these show indirect to direct
bandgap transition when the thickness is reduced to a monolayer from the bulk [8]. TMDCs with
monolayer thickness have attracted considerable attention due to their specific, electronic,
optoelectronic, spin and valley properties, making them a promising materials for high-
performance future electronic, optoelectronic, spintronics and valleytronics devices [6, 9-10].
MoSe2 is one of the crucial members of the TMDCs family. It has emerged as a promising
candidate for future electronic and optoelectronic applications due to the small direct bandgap,
high carrier mobility and high on-off ratio larger than 106 [4, 6]. Further, the smaller bandgap,
making the MoSe2 an exciting candidate for the IR light-emitting devices. Bulk (monolayer)
MoSe2 shows an indirect (direct) bandgap with a value of 1.1 eV (1.54 eV) [11]. The performance
of electronic and optical devices based on the MoSe2 and other these kinds of 2D materials will be
significantly influenced by the change of thermal properties such as thermal expansion coefficient
(TEC) and thermal conductivity of the materials, with temperature. Generally, 2D materials are
supported on some substrates for device applications, like SiO2/Si, which leads to induced strain
into the system due to the TEC mismatch between the MX2 and substrate. The developed strain
into the system significantly impacts the fundamental properties of these 2D materials [12-13].
Therefore, to increase the performance and reliability of electronic and optical devices based on
3
these TMDCs, it becomes pertinent to understand the behaviour of TEC as a function of
temperature and induced strain/stress due to TEC mismatch between the MX2 and substrate.
Raman spectroscopic technique has been proved to be very useful for probing 2D as well bulk
systems and their various aspects such as layer stacking geometry, strain effect, thermal properties,
defects, number of layers, and so on [5,7,12-20]. Several authors have employed temperature-
dependent Raman scattering to estimate thermal properties like TEC and thermal conductivity via
monitoring the behaviour of the phonon modes as a function of temperature [14-16]. Additionally,
anharmonicity present in the material affects the dynamics of the charge carriers via controlling
the strength of electron-phonon and phonon-phonon interactions, which may also significantly
impact the functioning of the devices. Electron-phonon coupling in 2D materials plays a crucial
role in controlling ballistic transport, Raman spectra, and dynamics of an excited state.
Anharmonicity, resulting from phonon-phonon interactions, and electron-phonon coupling may be
understood by monitoring the phonon modes' temperature-dependent behaviour.
Here, we report an in-depth temperature-dependent Raman study on layered MoSe2, grown by
chemical vapour deposition (CVD) method, in a wide temperature range of 4 to 330 K. The
measurements were done on both monolayer (1L) and bilayer (2L) MoSe2. We exracted the
thermal expansion coefficient by monitoring the temperature dependence of the first-order optical
phonon modes. So far, such studies on the MoSe2 have not been undertaken to the best of our
knowledge. Also, the temperature-dependent behaviour of the low-frequency interlayer modes,
forbidden Raman or IR active modes, and especially second and higher-order phonon modes are
not explored. In this paper, we have focused on these unexplored aspects of this material.
Interestingly, we observed the broadening of the1gA and 2
2gE shear phonon modes in the low
temperature window attributed to the electron-phonon coupling. We believe that our detailed
4
studies will pave the way for further studies on MoSe2 and other 2D materials in the future in this
direction.
2. Results and Discussions
2.1 Characterizations of MoSe2
Figure 1 (b) shows the optical micrograph of CVD grown triangular-shaped flake of the monolayer
MoSe2 on SiO2/Si substrate. Details about synthesis, Raman and photoluminescence (PL)
measurements could also be found in the supplementary part. Raman and PL identify flakes of
mono and bilayer (2L), and thickness is confirmed by atomic force microscopy (AFM). The
average thickness of the flake from the substrate is evaluated to be around 1.0 nm, which is close
to the value for the monolayer MoSe2 [4] (see Fig. 1 (c) ). Figure 1 (d) shows the room temperature
PL spectrum, we observed a strong PL emission from the corner, located at ~ 1.53 eV, consistent
with the excitonic transition band exA at the K point of the Brillouin zone (BZ) for the monolayer
MoSe2 [19, 21]. Surprisingly, moving from corner to center of the flake, the peak position of the
PL signal is red-shifted by a value of ~ 8 to10 meV and is located at 1.52 eV at the center of the
flake. We also noticed that the intensity of the PL signal in the central region is approximately two
times quenched in comparison to that of the corner region.
Figure 1 (e) shows the room temperature Raman spectrum of the MoSe2 flake. We find that the
spectra exhibit a strong peak at ~ 240.3 cm-1, and a very weak peak at ~ 286.1 cm-1 corresponds to
the Raman active out-of-plane (1gA ) and in-plane ( 1
2gE ) vibrational modes, respectively. It should
be noted that with changes in the number of layers for these 2D systems, symmetry representation
of phonon modes changes. We have used the terminology as used for the bulk system for both
monolayer and bilayer for convenience. The frequency difference between1gA and 1
2gE modes is
found to be ~ 45.6 cm-1 indicating the monolayer of MoSe2 flake [22]. We did not observe the
5
Raman peak associated with the interlayer interaction ( 2
2gE ) mode further suggesting that the flake
is a monolayer. AFM surface morphology and the Raman and PL intensity mapping are used to
check the uniformity of the MoSe2 flake. AFM surface morphology shows the remarkable
uniformity of the MoSe2 flake, see Fig 1(c). Figure 1 (f) illustrates typical PL mapping of the entire
triangular flake performed at room temperature, showing the PL intensity distribution of the
observed PL signal corresponding to theexA exciton. From Fig. 1 (f), we could see that the PL
intensity distribution is nearly uniform at the edges and corners of the MoSe2 flake, while a
significant quenching in intensity is observed at the center of the flake. Figure 1 (g-j) shows the
Raman intensity mapping performed at room temperature for the characteristic1gE ,
1gA , 1
2gE and
2
2uA modes from the entire flake of MoSe2, respectively. We observe that the Raman intensity
distribution of1gA and 1
2gE mode is nearly uniform across the whole flake, indicating the uniformity
of the MoSe2 flake. PL mapping yields more information about the uniformity and quality of the
samples than does Raman mapping because the PL emission is very sensitive to the thickness and
defects of the materials. In addition to the quenching in the PL band at the center, we also notice
the broadening (see Fig. 1 (d). The broad PL band at the center may result from defects that reduce
the lifetime of excitons, causing the broadening of the PL band. Meanwhile, defects may also offer
non-radiative channels, which may also quench the intensity of the PL band at the center.
Figure 2 (a) shows the optical micrograph of CVD grown MoSe2 flake consisting of the two
different thickness regions. We have selected two regions of interest from the flake exhibiting
layers of different thickness. Area represented by F1 and F2 exhibit monolayer and bilayer MoSe2,
respectively. Figure 2(b) depicts the room temperature Raman spectrum collected from the F1 and
F2 regions. The first-order Raman active 1gA and 1
2gE modes are observed at 240.1 cm-1, 241.1
6
cm-1; 285.6 cm-1, 284.9 cm-1 for F1 and F2 regions, respectively. The frequency difference between
1
2gE and 1gA modes is observed to be 45.5 cm-1 and 43.8 cm-1 for the F1 and F2 region, suggesting
the monolayer and bilayer MoSe2, respectively [22]. The1gA mode softens while 1
2gE mode stiffens
as we move from bilayer to monolayer region, which is in line with the previous reports [7, 22].
We notice that for the F1 region, no additional mode is observed in the low frequency (< 50) range.
But for the F2 region, we observed one additional mode, centered at ~ 17.4 cm-1 attributed to the
interlayer shear mode ( 2
2gE ) for the bilayer MoSe2 [23], which further indicates the F1 and F2
regions exhibit monolayer and bilayer thickness, respectively. Observed Raman modes near ~ 170
cm-1 and ~352 cm-1 have been assigned as1gE and 2
2uA modes, respectively, in earlier studies [19]
However, the peak observed near 352 cm-1 is also assigned as 1
2 2/g uB A [22-23], suggesting the
ambiguity in assigning proper symmetry to this mode. 1gE mode is Raman active but is normally
forbidden in backscattering Raman scattering measurements. While 2
2uA mode is infrared active
(IR) and is associated with the out-of-plane vibration of both the Mo and Se atoms, see inset in Fig
1(e). At room temperature, for the case of a monolayer, 1gE and 2
2uA modes are observed at 170.4
cm-1 and 352.8 cm-1, respectively. As the thickness changes from monolayer to the bilayer, 1gE
mode is strongly redshifted and is observed at 167.9 cm-1, while 2
2uA mode remains unchanged.
Moreover, both modes are prominent in bilayer as compared to in the case of the monolayer.
Figure 2 (c) shows the room temperature PL spectrum of the MoSe2 flake collected from both the
F1 (black) as well as F2 (red) regions. For the F1 area, we observed a very intense PL signal
centered at ~ 1.54 eV and is very close to exA for the monolayer MoSe2. As we move from F1 to
F2 region, PL signal is strongly quenched by ~ 40 times compared to that of F1 region, is centered
7
at ~ 1.48 eV and is redshifted by ~ 60 meV ( see bottom inset in Fig. 2 (c)). The quenching in
intensity and red-shift nature of the PL signal for the bilayer (F2 region) may be due to the
transition from direct bandgap in the monolayer to indirect bandgap in the bilayer MoSe2. Figures
2 (d-h) show the Raman intensity mapped of the first-order modes 2
2gE ,1gE ,
1gA , 1
2gE and 2
2uA ,
respectively of the entire flake. Raman intensity map of the 1gA mode demonstrates that the bilayer
MoSe2 regions give a higher Raman intensity than monolayer regions. In contrast, opposite
behaviour is observed for the case of 1
2gE mode. We also observed that the intensity distribution of
the modes is uniform across monolayer and bilayer region in the entire flake, reflecting the
uniformity of our CVD grown MoSe2. Moreover, the Raman intensity map of the 1gE and 2
2uA
modes shows that both these modes are significantly intense in bilayer compared to monolayer.
To confirm further, we also performed similar measurements on another triangular-shaped MoSe2
flake consisting of both monolayer and bilayer regions; for details, see Fig. S2 in supplementary
information and details therein.
2.2 Multi-phonons Raman scattering in MoSe2
Figures 3 (a) and 3 (b) show the Raman spectrum of monolayer and bilayer MoSe2 at 4 K, in a
spectral range of 10-640 cm-1, respectively. In the yellow shaded area, insets show the amplified
spectra in the spectral range of 80-200 cm-1, 260-500 cm-1 and 540-640 cm-1. The spectra are fitted
with a sum of Lorentzian functions to extract mode frequency ( ), full width at half maximum
(FWHM) and intensity of the individual mode. The observed modes are in excellent agreement
with the recently published reports on high-quality MoSe2 sample grown by vapour phase
chalcogenization and mechanical exfoliation method [19, 24]. For convenience, we have labelled
the observed modes as S1 to S20, and the observed modes and their corresponding symmetries are
listed in Table-I, for details about the phonon modes at the gamma point (see supplementary part).
8
The modes symmetry assignment is done according to the previous reports and our polarization-
dependent measurements. The interlayer mode 2
2gE (S1) is observed at 17.4 cm-1 for bilayer (absent
for 1L). The first-order modes1gA (S7) and 1
2gE (S8) are observed at 242.3 cm-1, 242.2 cm-1; 287.9
cm-1 285.9 cm-1 in monolayer, and bilayer. Forbidden first-order Raman active modes in
backscattering geometry 1gE (S6) and IR active 2
2uA (S12) are observed at 174.1 cm-1, 168.8 cm-1;
355.1 cm-1 354 cm-1 in monolayer and bilayer, respectively. In addition to the well-known first-
order optical modes, 2
2gE , 1gE ,
1gA , 1
2gE and 2
2uA , we also observe first-order longitudinal acoustic
( )LA , transverse ( )TA and out-of-plane ( )ZA modes near M or K symmetry points in the BZ along
with a large number of second- or higher-order phonon modes. The first-order LA mode near the
M point of the BZ is observed at 152.1 cm-1 and 147.3 cm-1 in monolayer and bilayer MoSe2,
respectively. Towards the low-frequency side of ( )LA M (S5) mode, two weak modes S2 and S3,
are observed at 124.8 (129.7) cm-1, and 121.9 (127.6) cm-1 in monolayer (bilayer); and are assigned
as TA and ZA , respectively, along M K direction in the BZ [19]. For the case of monolayer,
overtones and combinations of optical and acoustical phonon modes from the M symmetry point
of the BZ are observed at 140.7 cm-1 ( 1
2 ( )gE LA M ; S4), 305.5 cm-1 ( 2 ( )LA M ; S9), 364.3 cm-1 (
1 ( ) ( )gA M LA M ; S13), 414.6 cm-1 ( ( ) 2 ( )TA M LA M ; S14), 432.7 cm-1 ( 1
2 ( ) ( )gE M L A M ; S15),
457.2 cm-1 (3 ( )LA M ; S17), 569.7 cm-1 ( ( ) 3 ( )TA M LA M ; S18), 584.6 cm-1 ( 1
2 ( ) 2 ( )gE M LA M ; S19)
and at 598 .2 cm-1 ( 4 ( )LA M ; S20). The observed second- and higher-order phonon modes and their
corresponding symmetry assignment are given in Table-I for the case of the bilayer. Further, we
notice a mode S16 at ~ 444.6 (441.7) cm-1 in monolayer (bilayer) at 4K, while at room temperature,
this mode is observed at ~ 441.5 (339.1) cm-1 in monolayer (bilayer) MoSe2. The energy of this
9
mode is close to the sum of 2
2uA (S12) and LA (S5) modes from the M point of the BZ. Interestingly,
this mode is absent when the spectra are excited using a 632.8 nm laser (see Fig. S3).
To decipher the symmetry assignment and to understand the angle-dependent nature of the phonon
modes with respect to the incident photons polarisation direction, especially for 1gE (S6) and 2
2uA
(S12) modes near 170 cm-1 and 350 cm-1, respectively; we carried out detailed polarised Raman
scattering measurements for both monolayer and bilayer of MoSe2. The polarization-dependent
measurements were done by rotating the direction of the incident light with an angle ( ) by
keeping fixed the position of the sample and direction of the scattered light as described in Ref
[25-26]. Insets in the green shaded area, see Fig. 3 (a) and 3 (b), are the angular dependence of the
intensity polar plots of the modes 2
2gE (S1), 1gE (S6),
1gA (S7), 1
2gE (S8) and 2
2uA (S12) for both
monolayer and bilayer. The intensity of the 2
2gE (S1) mode shows isotropic nature with respect to
the polarisation angle, i.e. intensity is invariant with respect to the rotation of polarisation angle.
However, the intensity of the1gA mode shows two-fold symmetric nature, i.e. it has a maximum
intensity at both 00 and 0180 , while the intensity approaches zero at 090 and 0270 . The angular
dependence of the intensity of the mode 1gE (S6) is similar to 2
2gE mode (i.e. intensity is independent
of polarisation angle). The polarisation-dependent results discussed above could also be seen in
the 2D colour contour maps of the Raman intensity versus Raman shift and as a function of
polarisation angle, which are shown as insets for (a) 1gA and (b) 2
2gE and 1gE in Fig. 3 in the grey
colour area for monolayer and bilayer MoSe2, respectively. The intensity of the 1
2gE (S8) mode
showed slight dependence on the rotation angle. Its intensity is little more at 00 than that at 090 ,
while angular dependence of the intensity of the mode 2
2uA (S12) is similar to that of 1gA mode.
Based on our polarization-dependent Raman observations and IR reflectance observations from
10
literature [2], we attribute modes near ~ 170 cm-1 (S6) and ~ 350 cm-1 (S12) to the first-order
Raman active 1gE mode and IR active 2
2uA mode, respectively.
The observed variation in intensity as a function of polarization angle may be understood within a
semi-classical approximation. As the incident and scattered polarised light lie in the XY plane, the
unit vector associated with incident ( ie ) and scattered ( ˆse ) light of polarisation may be decomposed
as 0 0[cos ( ), ( ), 0)]sin and
0 0[cos ( ), ( ), 0)]sin , respectively, where 0 is an arbitrary
angle from the x-axis and varies from 00 to 0360 . Within the semi-classical approximation,
Raman scattering intensity of the first-order phonon modes is given as 2
intˆ ˆ| . . |t
s iI e R e , where R is the
Raman tensor [27-28]. Using the above expression and Raman tensor [28], the intensity of the1gE
,1gA and 1
2gE modes for our experimental geometry is given as 1
0gEI ,
1
2cosgAI a and
12
2 2 2(cos sin )gE
I d , respectively. Within the semi-classical approximation, the following
observations can be made: (i) Intensity of the1gE mode is zero, suggesting it should be absent in
the backscattering geometry. (ii) Intensity of the1gA mode is maximum when ie and ˆ
se are parallel
to each other, i.e. 00 , and it reduces to zero when 090 . (iii) Intensity of the 1
2gE mode remains
invariant with respect to rotation of polarisation angle. From our above discussions, we may
conclude that within the semi-classical approximation, the modes with E type symmetry are not
affected by polarisation configuration and is either observed or forbidden in both parallel and
cross-polarization configuration. However, modes with A type symmetry are strongly affected by
the polarisation configuration and can be observed only in parallel polarisation configuration.
Except for 1
2gE (S8), the solid lines are the fitted curves from the above expressions suggesting that
the experimental results are in very good agreement with the semi-classical approximation. The
11
intensity pattern of the mode 1
2gE in both monolayer and bilayer MoSe2, is slightly smaller in cross-
polarization as compared to parallel polarisation configurations, and forming a semi-lobe kind of
structure (see Fig. 3 (a) and 3 (b)). Intensity of the 1
2gE mode may be fitted well using the combined
functions i.e. 2 2 2 2 2( )cos sind a d , and we can see that the overall fitting is modest (see Fig. 3).
The observed anisotropic nature of 1
2gE mode may arise due the strong photon-electron-phonon
coupling, which may be understood within the quantum mechanical picture [26].
2.3 Thermal expansion coefficient and temperature-dependent frequency of the first-order
optical modes
To understand the temperature dependence of the phonon modes quantitatively, we extracted self-
energy parameters such as mode frequency ( ) and FWHM ( ) of the phonon modes using
Lorentzian functions fitting. Figure 4 (a) illustrates temperature dependence of the frequency of
the 1gA (S7) and 1
2gE (S8) phonon modes for monolayer MoSe2. Both these modes, i.e. 1gA and 1
2gE ,
stiffens with decreasing temperature down to ~ 80 K, and below this, a sudden rise (drop) in
frequency is observed for the case of 1gA ( 1
2gE ). On further cooling, till 4K, both these modes
remain nearly temperature independent. Figure 4 (b) illustrates temperature dependence of the
frequency of 1gA and 1
2gE phonon modes for bilayer MoSe2. Both 1gA and 1
2gE mode stiffens with the
decreasing temperature down to ~ 80 K; interestingly, a softening in frequency is observed for
both the phonon modes on further cooling. Figure 5 (a) shows the temperature dependence of the
frequency of 1gE (S6) and 2
2uA (S12) modes for monolayer. We observe that variations in the
frequency of 1gE and 2
2uA modes in temperature window 330 to ~ 80 K is normal, and at ~ 80 K, a
sudden drop in frequency is observed for the case of 1gE mode, and below 80 K, it again starts to
harden on further cooling till 4 K; while frequency of the 2
2uA mode remains nearly constant below
12
~ 80 K. Figure 5 (b) shows the temperature dependence of the frequency of the 2
2gE (S1), 1gE (S6)
and 2
2uA (S12) modes for bilayer. Temperature-dependent behaviour of the frequency of the mode
2
2uA (see Fig. 5(b)) is similar to that of 1gA and 1
2gE modes for bilayer (see Fig 4 (b)). The
temperature-dependent shift in the frequency of the 2
2gE and 1gE modes is surprisingly fascinating.
Since the overall change in frequency of the 2
2gE and 1gE modes is minor (~ 0.6 cm-1) nevertheless,
we could see apparent variations in the mode frequency with temperature, and this can be divided
into four regions: (i) Modes show hardening in the temperature range of 330-280 K, (ii) and then
shows softening in range of 280-200 K, (iii) afterwards it again show hardening from ~200 K to
~80 K, (iv) and with further decrease in temperature, it again shows softening till the lowest
recorded temperature (4 K). Details about the temperature dependence of the first-order acoustic
and second and higher-order phonon modes are given in supplementary information (please see
supplementary section 3).
The temperature-dependent shift in frequency of the phonon modes of the free-standing MoSe2
may be understood via: (i) anharmonic effect, which arises due to change in the self-energy
parameter because of phonon-phonon coupling (ii) quasi-harmonic effect which arises due to
thermal expansion of the lattice. The change in the phonon mode frequency as a function of
temperature considering the above two effects may be given as
( ) ( ) ( )Anh ET T T (1)
where ( )Anh T and ( )E T correspond to change in the mode frequency due to phonon-phonon
anharmonic effect and thermal expansion of the lattice, respectively. The first term in Eqn. 1 arises
due to a change in phonon self-energy because of the anharmonic effect and is given as [29]
2
1( ) (1 )xAnh eT A
, where / 2 Bx k T and A is a self-energy constant parameter, representing the
13
contributions from three phonon anharmonic effect. An optical phonon decays into two phonons
with equal frequencies and opposite momentum in the three-phonon anharmonic model.
Furthermore, with the variations in temperature, the lattice parameter of the MoSe2 would change
due to thermal expansion of the lattice parameter, resulting in a variation in the phonon mode
frequency as a function of temperature. The contribution of thermal expansion effect to the shift
in phonon mode frequency is given as [30] 2
0
0 0( ) exp[ 3 ( ) ]
T
E MoSe
T
T T dT , where is the
Gruneisen parameter of a particular mode and 2( )MoSe T is temperature-dependent TEC of MoSe2.
We note that both positive and negative TEC has been reported for MoSe2 [31-33]. In the present
case, MoSe2 is not free-standing but supported by SiO2/Si substrate, and SiO2 has negative
(positive) TEC at low (high) temperature [34]. Therefore, in addition to the mentioned two effects
above (i.e. anharmonic and thermal expansion effects), thermally induced strain results from TEC
mismatch between MoSe2 and substrate should also be considered to understand the net change in
the phonon mode frequency with temperature. Change in the phonon mode frequency, considering
these effects, as a function of temperature is given as [15, 29-30]
( ) ( ) ( ) ( )Anh E sT T T T (2)
The last term ( )s T is the change in the mode frequency corresponding to the strain effect due to
TEC mismatch. It can be expressed as [15]2 2
0
( ) ( ) [ ( ) ( )]
T
S SiO MoSe
T
T T T T dT , where is the
strain coefficient of a particular mode and 2( )SiO T is the temperature-dependent TEC of SiO2. To
estimate2( )MoSe T , we have used the value of ( )
as
1
13.7 / %gA cm and 1
2
11.2 / %gE
cm [12].
TEC of SiO2 was taken from [34] and was integrated out while estimating TEC for MoSe2. The
14
product of the mode Grüneisen parameter and thermal expansion coefficient may be expressed by
a polynomial of temperature and is given as
2
2
0 1 2( )MoSe T p p T p T (3)
where p0, p1 and p2 are the constant parameters, whose values are obtained as a fitting parameter
by the best fit to the temperature dependence of the frequency of the modes. To extract the TEC
of MoSe2, we have fitted the frequency of1gA and 1
2gE mode in the temperature range of 80-330 K
for monolayer and bilayer using the Eqn. 2 and Eqn. 3. The above Eqn. 3 can be used to estimate
TEC as a function of temperature. We have adopted the theoretically calculated value of the mode
Gruneisen parameter (~ 1.7 for1gA and ~ 0.8 for 1
2gE ) to estimate the TEC of MoSe2 [32] in both
out-of-plane and in-plane direction. Figures 4 (c) and 4 (e) show the temperature-dependent TEC
of the in-plane ( 1
2gE ) and out-of-plane (1gA ) modes for monolayer and bilayer MoSe2, respectively.
Our temperature-dependent results show that the TEC of1gA mode for both system (1L and 2L
MoSe2) decreases sharply with decreasing temperature starting from 330 to ~200 K; on further
cooling, a mild decrease in TEC is observed. While TEC corresponding to the 1
2gE mode decreases
with decreasing temperature from 330 to ~200 K, and surprisingly below 180 K, an increase is
seen with further lowering the temperature (see shaded area). We also estimated the volumetric
TEC given as 2 a c [32], where a and c are the linear TEC for in-plane and out-of-plane
direction. Figure 4 (d) and 4 (f) shows the volumetric TEC as a function of temperature. Room
temperature linear TEC corresponding to the 1
2gE and1gA modes together with volumetric TEC are
listed in Table-II. The volumetric TEC values at room temperature are found to be 21.4 ×10-6 K-1
and 23.5×10-6 K-1 for monolayer and bilayer MoSe2, respectively, see Table-II. The estimated
volumetric TEC results are in excellent agreement with previously reported values for monolayer
15
MoSe2 supported by SiO2/Si substrate [36]. Our estimated linear in-plane TEC value for the
monolayer is very close to the theoretically calculated TEC for monolayer MoSe2 [33], while it is
nearly ten times smaller than that of experimentally reported in-plane TEC for free-standing
monolayer MoSe2 [37]. Furthermore, TEC corresponding to the1gA mode is larger than that of 1
2gE
mode for monolayer, which is in line with earlier reports on other TMDCs [35, 38]. The larger
TEC of 1gA mode may be understood as: in layered materials, the out-of-plane direction is confined
weekly compared to the in-plane direction; therefore, it is easier to deform the out-of-plane
direction than the in-plane direction [39]. With the change in thickness from monolayer to the
bilayer, we observed decreased TEC of1gA mode, reflecting the increase in mode strength. For
bilayer, an increment in TEC of 1
2gE mode is observed compared to that of the monolayer, which
differs from the earlier report on MoSe2 [37]. Therefore, further theoretical and experimental
studies on these 2D materials are required to decipher discrepancies in TEC, especially for in-
plane linear TEC.
Now we will focus on the observed anomalies in the mode frequencies at low temperature. The
observed kink in frequencies of the modes at low temperature for monolayer, as shown in Fig. 4
(a) and Fig. 5 (a), may be due to induced strain owing to TEC mismatch between MoSe2 and the
substrate. The induced strain due to TEC mismatch may affect the weak van der walls forces,
leading to the slippage, realignment or change in surface morphology of MoSe2 films on the
substrate and forming of wrinkles or ripples into the system, which may affect the frequency of
the modes. Similar anomalies are also observed in the case of other MX2 systems [38, 40]. For
bilayer, we observed a decrease in the mode frequency (see Fig 4(b) and Fig 5 (b)) below ~ 100
K. A negative TEC has been reported for MoSe2 at low temperature [33], and this may generate
the tensile stress into the system at low temperature, which may give rise to the anomalous decrease
16
in the mode frequency for bilayer at low temperature. To confirm the realignment of MoSe2 films,
we did AFM characterizations after the performance of temperature-dependent Raman
measurements of the same flake as optically shown in Fig. 2 (a). Figure 2 (i) shows the AFM
image, and the insets depict the step height profile. The step height profile along the white line
shows the height of monolayer MoSe2 from the substrate. We notice that the average thickness of
monolayer (F1 region) from the substrate is observed to be ~ 3 nm which is significantly larger
than that of the actual thickness of monolayer MoSe2 from substrate reflecting the slippage or
realignment of MoSe2. The step height profile along the blue line shows the height of the bilayer
along with monolayer MoSe2 from the substrate. The average thickness of bilayer from the
substrate is ~ 2 nm, reflecting a very weak deviation from the reported thickness for the bilayer
MoSe2 from the substrate, suggesting the weak effect of induced strain due to TEC mismatch
between bilayer and substrate. It is in line with the fact that induced strain due to TEC mismatch
gradually decreases with the increasing number of layers and becomes negligible for the bulk.
2.4 Electron-Phonon coupling and lifetimes of the first-order optical modes
A strong electron-phonon coupling, which limits the electronic mobility of semiconductors, can
significantly affect the self-energy parameters of the phonon modes, and this effect may be
captured via a detailed temperature-dependent Raman measurement. The temperature-dependent
mobility in semiconductor MoSe2 was attributed to the scattering of carriers by optical phonons
which corresponds to the fluctuations of the layer thickness [41], implying that the A1g, phonon
with atomic displacements along c-axis, and 2
2gE phonon, interlayer shear mode, modes may be
involved in controlling the mobility of the carriers. In this section, we focus our attention on the
temperature dependence of the FWHM of the phonon modes and the role of electron-phonon
coupling in the temperature-dependent evolution of FWHM. In a high-quality sample, FWHM of
17
the phonon modes at finite temperature may be affected by the contributions of two factors: (i)
phonon-phonon coupling (-ph ph ) (ii) electron-phonon interactions (
-e ph ). Therefore, considering
these factors, the temperature dependence of the FWHM of the phonon modes may be given as
[42-43]
- -( ) ( ) ( )ph ph e phT T T (4)
The first term - ( )ph ph T arises from decaying an optical phonon into two phonons of the same energy
and opposite momentum satisfying the energy and momentum conservation rules. In general,
FWHM of the phonon mode increases as temperature increases. With increasing temperature,
populations of the phonons also increase as the lifetime of the phonons is inversely proportional
to the FWHM; as a result, significant broadening (increase) in FWHM is expected with increasing
temperature. The contributions from the three-phonon anharmonic effect to the FWHM of phonons
as suggested by Klemens may be expressed as [29] 2
1( ) (0) (1 )xph ph ph ph eT C
, where / 2 Bx k T
C is a constant parameter and (0)ph ph is the FWHM at 0 K, and the term ( )ph ph T
is expected to
dominate at high temperature. The second term - ( )e ph T arises as a result of contributions from
the electron-phonon interactions. Temperature-dependent contribution from the electron-phonon
coupling interactions to the FWHM may be given as [42-43] 1 1
1 1( ) (0)[( ) ( )]x xe ph e ph e eT
, where
/ 2 Bx k T and (0)e ph is the FWHM resulting from the electron-phonon coupling effect at 0 K,
and the term - ( )e ph T is expected to dominate at low temperature. The expression for ( )e ph T
represents the difference in occupations of states below and above the Fermi energy level and may
be used to understand the temperature-dependent shift in the FWHM of the phonon modes.
Occupations of the filled states below Fermi level decreases with an increase in temperature, while
empty states above Fermi level are occupied more and may result in narrowing (broadening) of
18
the FWHM with increasing (decreasing) temperature [42]. Here, a renormalization of the phonon
modes may be understood due to phonon induced electron-hole pair creations. With increasing
temperature, empty states above the Fermi level starts filling up and this blocks the generation of
the phonon induced electron-hole pairs and hence affects the phonon self-energy. In particular, it
is expected that at low (high) temperature the phonon lifetime will be less (more) and as a result
linewidth will be more (less); based on the pure electron-phonon coupling effect one expects that
linewidth will be more at low temperature and less at a higher temperature. Also, the bandgap in
monolayer MoSe2 is significantly higher than that in the bilayer; therefore, the effect of electron-
phonon coupling is expected to be more visible in bilayer owing to the reduced bandgap.
Figures 6 (a) and 6 (b) show the temperature dependence of the FWHM of the 1gA and 1
2gE modes
for monolayer and bilayer MoSe2, respectively. For monolayer, we could see that FWHM of the
1gA mode decreases with decreasing temperature showing normal temperature dependence; on the
other hand, for 1
2gE mode, we did not observe any apparent variations in FWHM with temperature
for both mono and bilayer. For bilayer, FWHM of the1gA mode shows normal temperature-
dependent behaviour in the temperature range of 330 to ~ 100 K, and surprisingly below 100 K,
an increase in FWHM is observed with further cooling till the lowest recorded temperature (4 K).
Figure 5 (c) shows the temperature dependence of the FWHM of the 1gE and 2
2uA modes for
monolayer MoSe2. We notice that the FWHM of the1gE mode decreases with decreasing
temperature, while the FWHM of the 2
2uA mode shows a non-monotonic trend with temperature.
Figure 5 (d) shows the temperature dependence of the FWHM of the 2
2gE , 1gE and 2
2uA modes for
bilayer. FWHM of the 2
2gE mode shows normal temperature-dependence; it decreases with
decreasing temperature, from room temperature to ~ 120 K. Quite surprisingly, on further lowering
19
the temperature, it starts to increase till 4 K attributed to the strong electron-phonon coupling. We
note that similar behaviour is also observed for the shear modes in Graphene [43]. FWHM of the
1gE mode shows normal temperature dependence till ~ 100 K, and it increases slightly below 100
K. On the other hand, the FWHM of 2
2uA mode shows normal temperature dependence in the entire
temperature range.
FWHM of the modes as a function of temperature are fitted using the above Eqn. 4, the solid red
lines in Fig. 5 (c and d) and Fig. 6 (a and b) are the fitted curves, and it is in very good agreement
with our experimental data. For the monolayer, fitting parameters (0)ph ph and (0)e ph
of the 1gA
mode are 1.4 cm-1 and 0.02 cm-1, respectively, suggesting that the dominating factor is the phonon-
phonon anharmonic effect. However, for the case of bilayer (0)ph ph and (0)e ph
of the 1gA mode
obtained from the fitting are 0.6 cm-1 and 0.7 cm-1, respectively, suggesting a significant role of
election-phonon coupling at low temperature, and which may also explain the observed increase
in the FWHM with decreasing temperature below 100 K. Overall -e ph is substantial in comparison
to ph ph
for bilayer system as anticipated earlier. The fitting parameters obtained from the FWHM
for other first-order optical phonon modes are summarised in Table-III. Our observation of an
anomalous increase in FWHM at low temperature, especially in the bilayer, deviating from normal
temperature behaviour, may be understood keeping the finite role of electron-phonon coupling in
these systems.
2.5 Infrared, Forbidden Raman and higher-order phonon modes
In addition to the first-order phonon modes, we observed a large number of phonon modes from
the high symmetry point of BZ. Further, we also observed first-order phonon modes, which are
either forbidden in backscattering geometry or IR active. The appearance of these forbidden and
20
IR active phonons from the BZ center and multi-phonon Raman scattering from other parts of the
BZ may be understood via resonance effect [22, 44], Fröhlich mechanism of exciton-phonon
coupling [45], and cascade theory of inelastic light scattering [46]. The resonance effect occurs
when the laser excitation energy is close to the excitonic energy state. As a result, under the
resonance condition, in addition to first-order Raman active phonon, first-order Raman active but
backscattering forbidden and IR active modes from BZ center as well as first-order acoustic
phonon modes from the high symmetry point of the BZ could also be observed. Furthermore, the
resonance excitation also gives rise to intense and enriched Raman spectrum, including second
and high-order phonon modes from the high symmetry point of the BZ [44]. Recently, Bilgin et
al. [19] theoretically estimated the energy of C exciton to be 2.33 eV, while experimentally, this
peak is reported at ~ 2.5 eV [47]. In our case, excitation wavelength 532 nm (2.33 eV) is near C
exciton energy; and could be expected to be a good condition for near resonance Raman scattering
effect. Under the resonance condition, the observation of backscattering forbidden (1gE ) and IR-
active ( 2
2uA ) phonons may be understood by the Fröhlich mechanism of exciton-phonon coupling.
Fröhlich mechanism of the exciton-phonon coupling may give rise to finite intensity of the
forbidden phonons proportional to (aq) 2, where a is Bohr radius of excition, and q is wave vector
of phonon. When the Bohr radius is much larger than the lattice parameter, then the forbidden
phonon modes may appear in Raman spectra under resonance condition. These conditions may be
satisfied easily in TMDCs, where the Bohr radius of the exciton is larger than the lattice parameter
[48]. As a result, they can couple to the phonons and give into a finite intensity even for the
forbidden phonons. It was recently reported that 1gE and 2
2uA modes are observed only for excitation
energies above 2.2 eV, and they both become intense close to the C exciton [24]. To confirm, we
excited the spectra with 632.8 nm (1.96 eV) laser; surprisingly, both 1gE and 2
2uA modes were not
21
seen in the Raman spectrum, see Fig. S3. Therefore, the appearance of the 1gE and 2
2uA modes in
our Raman spectrum of the MoSe2 using 532 nm laser reflects the resonance effect with C exciton.
Further, it should be noted that we observed intense phonon modes up to fourth-order overtone
4 ( )LA M , and the appearance of high-order phonon modes may be understood by considering the
cascade process of the Raman scattering of light [46]. In cases when the energy of the incident
photons is less than that of the bandgap energy of the material, i.e. (i gE E ), the intensity of the
nth-order phonon modes varies as ng , where g is the typical electron-phonon coupling constant
which is generally much less than unity, n is an order of the phonon modes. Therefore, the intensity
of higher-order modes decrease extremely fast as n increase, and the higher-order modes are
generally expected to be very weak or absent in the Raman spectrum. When the energies of both
incident and scattered photons are above the bandgap, i.e. ,i s gE E E , inelastic scattering of light
occurs via cascade process. In such a case, the intensity of modes are independent of electron-
phonon coupling and depends mainly on the dispersion curves of the electron and hole bands; as
a result, higher-order phonon modes may appear in the Raman spectrum. We note that we have
used a 532 nm (2.33 eV) laser as excitation energy which is above the bandgap energy ( gE ~ 1.55
eV) [11]. Therefore above discussed cascade process of Raman scattering may be easily satisfied,
which may give rise to intense high-order phonon modes in the Raman spectrum in our case.
3. Conclusion
In conclusion, we performed a comprehensive temperature- and polarisation-dependent Raman
study on CVD grown MoSe2 supported by SiO2(~ 300 nm)/Si in a wide temperature and broad
spectral range of 10-700 cm-1. A large number of phonons modes were observed, up to fourth-
order as well forbidden Raman and IR modes, understood by considering the resonance effect,
Fröhlich mechanism of exciton-phonon coupling and cascade theory of inelastic light scattering.
22
The thermal expansion coefficient is extracted for both mono and bilayer MoSe2 as a function of
temperature, and the effect of induced strain from the underlying substrate is found to be significant
for the case of a monolayer. The observed temperature evolution of the linewidth of the 1gA and
2
2gE mode suggests that electron-phonon processes are involved in addition to the phonon-phonon
anharmonicity, and is found to be dominating in the case of the bilayer.
Acknowledgement: PK acknowledges the Department of Science and Technology (DST) and IIT
Mandi, India, for the financial support.
References:
[1] T. Sekine et al., Solid State Commu. 35, 371 (1980).
[2] T. Sekine et al., J. Phys. Soc. Jpn. 49, 1069 (1980).
[3] K. S. Novoselov et al., Science 306, 666 (2004).
[4] X. Wang et al., ACS Nano 8, 5125 (2014).
[5] A. A. Puretzky et al., ACS Nano 9, 6333 (2015).
[6] S. Larentis et al., App. Phy. Lett. 101, 223104 (2012).
[7] H. Li, et al., Adv. Funct. Mater. 22, 1385 (2012).
[8] A. Splendiani et al., Nano Lett.10, 1271(2010).
[9] J. Isberg et al., Nat. Materials 12, 760 (2013).
[10] N. Kumar et al., Nanoscale 6, 12690 (2014).
[11] S. Tongay et al., Nano Lett.12, 5576 (2012).
[12] M. Yagmurcukardes et al., Phys. Rev. B 97, 115427 (2018).
[13] E. Blundo et al., Phys. Rev. Res. 2, 012024 (2020).
[14] D. Yoon et al., Nano Lett.11, 3227 (2011).
[15] S. Linas et al., Phys. Rev. B 91, 075426 (2015).
[16] R. Yan et al., ACS Nano 8, 986 (2014).
[17] P. Kumar et al., App. Phys. Lett. 100, 222602 (2012).
[18] P. Kumar et al., J. Phys.: Condens. Matter 26, 305403 (2014).
[19] I. Bilgin et al., ACS Nano 12, 740 (2018).
23
[20] B. Singh et al., J. Phys.: Condens. Matter 31, 065603 (2019).
[21] P. Tonndorf et al., Opt. Exp. 21, 4908 (2013).
[22] K. Kim et al., ACS Nano 10, 8113 (2016).
[23] X. Lu et al., Adv. Mater. 27, 4502 (2015).
[24] P. Soubelet et al., Phys. Rev. B 93, 155407 (2016).
[25] B. Singh et al., Phys. Rev. Res. 2, 023162 (2020).
[26] D. Kumar et al., J. Phys.: Condens. Matter 32, 415702 (2020).
[27] Light Scattering in Solid II, edited by M. Cardona and G. Guntherodt, Springer Verlag
Berlin (1982).
[28] R. Loudon, Adv. Phys. 50, 813 (2001).
[29] P.G. Klemens, Phys. Rev. 148, 845(1966).
[30] J. Menendez and M. Cardona, Phys. Rev. B 29, 2051 (1984).
[31] S. H. El-Mahalaway and B. L. Evans, Appl. Cryst. 9, 403 (1976).
[32] Y. Ding and B. Xiao, RSC Adv. 5, 18391 (2015).
[33] C. Sevik, Phys. Rev. B 89, 035422 (2014).
[34] Standard Reference Material 739 Certificate; National Institute of Standards and
Technology, 1991.
[35] X. Huang et al., Sci. Rep. 6, 32236 (2016).
[36] M. Yang et al., App. Phy. Lett. 110, 093108 (2017).
[37] X. Hu et al., Phys. Rev. Lett. 120, 055902 (2018).
[38} L. Su et al., Nanoscale 6, 4920 (2014).
[39] C. K. Gan and Y. Y. F. Liu, Phys. Rev. B 94, 134303 (2016).
[40] D. Kumar et al., J. Phys.: Condens Matter 31, 505403 (2019).
[41] R. Fivaz and E. Mooser, Phys. Rev. 163, 743 (1967).
[42] N. Bonin et al., Phys. Rev. Lett. 99, 176802 (2007).
[43] C. Cong and T. Yu, Nat. Commun. 5, 4709 (2014).
[44] D. Kumar et al., Nanotechnology. 32, 285705 (2021).
[45] R. M. Martin and T.C. Damen, Phys. Rev. Lett. 26, 86 (1971).
[46] R.M. Martin and C.M. Varma, Phys. Rev. Lett. 26, 1241 (1971).
[47] H.G. Park et al., Sci. Rep. 8, 3173 (2018).
[48] D.V. Tuan et al., Phys. Rev. B 98, 125308 (2018).
24
Table-I: List of the experimentally observed modes along with their symmetry assignments and
frequency at 4 K for the 1L and 2L MoSe2. Units are in cm-1.
Mode assignment
Frequency ( )
1L 2L
S1 [ 2
2 ( )gE ]
-
17.8±0.02
S2 [ ( )TA M K ] 124.8±0.8 121.9±0.3
S3 [ ( )ZA M K ] 129.7±0.5 127.6±0.4
S4 [ 1
2 ( )gE LA M ] 140.7±0.8 137.8±0.3
S5 [ ( )LA M ] 152.1±0.3 147.3±0.3
#1 160±1.0 -
S6 [1 ( )gE ] 174.1±0.6 168.8±0.4
S7 [1 ( )gA ] 242.3±0.1 242.2±0.02
S8 [ 1
2 ( )gE ] 287.9±0.2 285.9±0.2
#2 298.0±0.5
S9 [ 2 ( )LA M ] 305.5±0.3 302.8±0.4
#3 312.7±0.8 - S10 321.2±0.1 318.9±0.2 S11 - 342.5±0.5
S12 [ 2
2 ( )uA ] 355.1±0.4 354.0±0.03
S13 [1 ( ) ( )gA M LA M ] 364.3±1.0 362.8±0.7
#4 387.4±2.5 389.2±2.5
S14 [ ( ) 2 ( )TA M LA M ] 414.6±1.2 411.4±0.6
S15 [ 1
2 ( ) ( )gE M LA M ] 432.7±0.3 430.2±0.2
S16 2
2 ( ) ( )uA M LA M 444.6±0.2 441.7±0.3
S17 [3 ( )LA M ] 457.2±0.1 453.9±0.1
S18 [ ( ) 3 ( )TA M LA M ] 569.7±0.4 567.0±0.4
S19 [ 1
2 ( ) 2 ( )gE M LA M ] 584.6±0.2 581.6±0.1
S20 [ 4 ( )LA M ] 598.2±0.1 595.4±0.4
25
Table-II: Room temperature linear and volumetric thermal expansion coefficients (TEC) for 1L
and 2L MoSe2 extracted using 1
2gE and 1gA modes. Units are in 10-6 K-1.
MoSe2 1
2( )gE 1( )gA 2 a c
1L 4.1 13.2 21.4
2L 9.1 5.3 23.5
Table-III: List of the fitting parameters obtained from FWHM of the first-order optical phonon
modes for 1L and 2L MoSe2. Units are in cm-1.
Electron-phonon + phonon-phonon coupling model
Modes
1L 2L
C
(0)ph ph
(0)e ph
C
(0)ph ph
(0)e ph
2
2 ( )gE - - - 0.007±0.001 1.1±0.04 1.3±0.3
1 ( )gE 1.1±0.3 3.5±1.5 0.06±1.5 0.9±0.2 2.8±0.9 1.0±0.9
1 ( )gA 0.2±0.1 1.4±0.02 0.02±0.01 0.3±0.02 0.6±0.1 0.7±0.08
1
2 ( )gE - - - - - -
2
2 ( )uA - - - 0.5±0.3 1.9±0.8 0.1±0.6
26
FIGURES:
FIGURE 1: (a) Schematic diagram of CVD growth MoSe2. (b) Optical and (c) AFM image of the
triangular 1L MoSe2 flake. (d) PL spectrum collected from the corner and center of MoSe2 flake
(e) Room temperature Raman spectrum. Insets show the ball stick model representing the
vibrations of the modes. (f) PL intensity mapping image corresponding to the PL emission
excitonic band ( exA ). (g-j) Raman intensity mapping images of the 1gE ,1gA , 1
2gE and 2
2uA phonon
modes.
27
FIGURE 2: (a) Optical micrograph of the 1L and the 2L thickness regions of the MoSe2 flake.
The areas of focus are indicated by F1 and F2. Room-temperature (b) Raman and (c) PL spectrum,
collected from the F1 (black) and F2 (red). The inset spectrum shows the PL peak position
collected from the F2 and, the inset image illustrates the PL intensity mapping image of the exA
excitonic band. (d-h) Raman intensity map of the 2
2gE ,1gE ,
1gA , 1
2gE and 2
2uA modes. (i) AFM image.
28
FIGURE 3: Raman spectra of 1L (a) and 2L (b) MoSe2 at 4 K. Insets in the yellow shaded area
show the amplified spectra. Insets in the green shaded area show the intensity polar plot of the 2
2gE
, 1gE ,
1gA , 1
2gE and 2
2uA modes; solid red lines are the fitted curves as described in the text. Insets
in grey colour area illustrate the 2D colour contour maps of the Raman intensity versus Raman
shift and as a function of polarisation angle for 1L and 2L MoSe2.
29
FIGURE 4: (a) and (b) Temperature dependence of the frequency of 1gA and 1
2gE modes for 1L
and 2L MoSe2, respectively. Solid red lines are the fitted curves as described in the text, and the
solid green lines are a guide to the eye. The shaded part illustrates the region where the mode
frequency shows anomalous behaviour. (c) and (e) Linear TEC corresponding to the 1gA (red) and
1
2gE (blue) modes and (d) and (f) Volumetric TEC in the temperature range of 80-330 K for 1L
and 2L MoSe2, respectively.
30
FIGURE 5: Temperature dependence (a) frequency, and (c) FWHM of 1gE and 2
2uA modes for
1L. Temperature dependence (b) frequency, and (d) FWHM of 2
2gE , 1gE and 2
2uA modes for 2L.
Solid red lines are the fitted curves described in the text, and solid green lines are guide to the eye.
Inset black and red plots in (d) describe individual contributions to the linewidth from phonon-
phonon and electron-phonon coupling, respectively.
31
FIGURE 6: Temperature dependence FWHM of 1gA and 1
2gE modes for (a) 1L; (b) 2L. Solid
red lines are the fitted curves described in the text, and solid green lines are guide to the eye. Inset
black and red plots in (b) describe individual contributions to the linewidth from phonon-phonon
and electron-phonon coupling, respectively.
32
Supplementary Information:
Electron-Phonon Coupling, Thermal Expansion Coefficient, Resonance Effect
and Phonon Dynamics in High Quality CVD Grown Mono and Bilayer MoSe2
Deepu Kumar1#, Vivek Kumar1, Rahul Kumar2, Mahesh Kumar2, Pradeep Kumar1*
1School of Basic Sciences, Indian Institute of Technology Mandi, 175005, India
2Department of Electrical Engineering, Indian Institute of Technology Jodhpur, 342001, India
#E-mail: [email protected]
*E-mail: [email protected]
1. Synthesis of MoSe2 using CVD
The triangular-shaped 2D MoSe2 was synthesized using a three zones thermal chemical vapour
deposition (CVD) system equipped with a quartz tube (2-inch diameter) under atmospheric
pressure. Figure 1 (a) (see in main text) is illustrating a schematic diagram of the CVD system
and showing that the two solid precursors, selenium (Se) and molybdenum trioxide (MoO3), were
loaded in the first and second zone of the CVD system, respectively. In this typical synthesis, 1.0
g of Se powder (99.9%, Sigma-Aldrich) and 0.04 g of MoO3 powder (99.9%, Sigma-Aldrich) were
placed 15 cm away from each other, and a cleaned SiO2/Si substrate was put face down on MoO3.
Before the chemical reaction, the tube was evacuated by a rotary pump and purged several times
with ultra-high pure Ar gas to remove oxygen and other contamination. After achieving
atmospheric pressure with purging Ar gas, the temperature of the MoO3 in the downstream zone
was increased up to 850°C with a heating rate of 20°C per min and the Se in the upstream zone
was heated at 300°C. The chemical reaction was performed for 30 min at above temperature values
in the presence of carrier gases’ mixture (1% H2 + 99% Ar) with a flow of 60 SCCM under
33
atmospheric pressure. Finally, the furnaces of the zones were switched off, and the tube was cooled
naturally back to room temperature.
2. Measurement techniques and Raman Phonon modes
The sample quality and crystal structure were assessed using Raman and PL features and the
intensity mapping at room temperature. Raman and photoluminescence (PL) measurements were
performed with Horiba LabRAM HR evolution in backscattering geometry. A 532 nm (2.33 eV)
laser was used to excite the Raman and PL spectra. Laser power was set to be very low ~ 0.2 mW
to avoid any local heating, as well as damage to the sample. The laser beam was focused on the
sample using a 100x long working objective lens, and the same objective lens was used to collect
scattered light from the sample. The scattered light from the sample was detected by using 1800
and 600 grating coupled with Peltier cooled Charge Coupled Device (CCD) detector for Raman
and PL measurements, respectively. For polarized Raman scattering, a polarizer and an analyzer
have been inserted into the incident light path on the sample and scattered light from the sample,
respectively. The polarization-dependent Raman measurements have been performed at room
temperature with rotating the direction of incident light with an interval of 200, while keeping fixed
the direction of scattered light and position of the sample. For temperature-dependent Raman
measurements, a 50x long working objective lens was used both to focus the incident beam on the
sample and to collect the scattered light. The temperature-dependent Raman measurements were
carried out using a closed cycle refrigerator (Montana Cryostat ) in a wide temperature range from
4 to 330 K , with an interval of 10 K and temperature accuracy of ± 0.1 K and waiting time for
each Raman measurement was ~ 10 minutes for better temperature stability.
Bulk MoSe2 belongs to the point group 4
6hD ( space group36 / ,#194)P m mc , the unit cell is
composed of two atomic formula (Z=2) with 6 atoms, which results in 18 phonon branches at the
34
point of the BZ, and these phonon branches can be expressed by following irreducible
representation as 1 2 2 1 1 1 2 22 2 2 2g u g u g u g uA A B B E E E E [1]. As the thickness of the material
decreases from bulk to few layers, resulting in symmetry variations due to the loss of the
translational symmetry perpendicular to the basal plane. For example, monolayer exhibiting the
non-centrosymmetric, and belonging to the point group 1
3hD (space group 6 2, #187)P m , there are 3
atoms per unit cell, giving into 9 phonon branches with the irreducible representation
' ' '' ''
1 22 2A E A E . The bilayer exhibits the centrosymmetric nature belonging to the point group
3
3dD (space group 3 1, #164)P m ). There are 6 atoms per unit cell, giving 18 phonon branches with an
irreducible representation1 23 3 3 3g g u uA E A E . The phonon modes with symmetry E are the in-
plane vibration of atoms and are doubly degenerate, while the phonon modes with symmetry A and
B correspond to the non-degenerate out-of-plane vibrations of atoms.
FIGURE S1: (a) Optical micrograph of CVD grown MoSe2 flake on SiO2/Si substrate consisting
of the monolayer (1L) and bilayer (2L) thickness regions (c) Atomic Force Microscope (AFM)
image. (c) Height profile taken along the green line for 1L from the substrate. (d) Height profile
taken along the blue line for 2L from 1L.
35
FIGURE S2: (a) Room-temperature Raman spectrum in the spectral range of 5-500 cm-1 collected
from 1L(black) 2L (red) by using a 532 nm laser. (b-f) Raman intensity mapping images of the 2
2gE
,1gE ,
1gA , 1
2gE and 2
2uA phonon modes, respectively.
FIGURE S3 Room-temperature Raman spectrum in the spectral range of 5-500 cm-1 collected
from 1L (black) 2L (red) by using a 632.8 nm laser.
3. Temperature dependence of the first-order acoustic and second and higher-order modes
Figure S4 (a) and S4 (b) Temperature evaluation of the Raman spectra in the temperature range of
4-330 K for 1L and 2L MoSe2, respectively. Figures S5 (a) and S5 (b) illustrate the temperature
dependence of the mode frequency ( ) and FWHM of the S3 and S5 phonon modes for the
36
monolayer and bilayer MoSe2, respectively. Following observations can be made from our
temperature-dependent Raman measurements: (i) For monolayer, frequency of both modes show
normal behaviour i.e. mode hardening with decreasing temperature; on the other hand, for bilayer,
we observe the anomalous behaviour of both modes i.e. mode softening with decreasing
temperature. (ii) For monolayer, FWHM of S3 mode show normal behaviour, i.e. FWHM
decreases with decreasing in temperature, while FWHM of S5 mode show normal behaviour down
to ~250 K with decreasing temperature, and below 200 K it exhibits nearly temperature-
independent behaviour till 4 K. For bilayer, FWHM of the S3 (S5) modes show normal
(anomalous) temperature-dependent behavior.
FIGURE S4: (a) and (b) Temperature evaluation of the Raman spectra in the temperature range
of 4-330 K for 1L and 2L MoSe2, respectively. Insets (a) show the expanded region of the 1gA (S7)
and 1
2gE (S8) modes for monolayer; and insets (b) show the expanded region of the 2
2gE (S1), 1gA
(S7) and 1
2gE (S8) modes for bilayer.
37
FIGURE S5: Temperature dependence of the frequency and FWHM of the S3 and S5 modes for
(a) 1L (b) 2L. Solid red lines are the fitted curves via a three-phonon anharmonic model described
in the text, and solid green lines are guide to the eye.
FIGURE S6: Temperature dependence of the phonon mode frequencies and FWHM of the modes
S9, S10, S15, S16 S17, S18, S19 and S20 for 1L MoSe2. Solid red lines are the fitted curves via
three phonon anharmonic model as described in the text and solid green lines are a guide to the
eye.
38
Figure S6 illustrates the temperature dependence of mode frequency ( ) and FWHM of S9, S10,
S15, S16 S17, S18, S19 and S20 phonon modes for the monolayer. Following observations can be
made: (i) Frequency of all phonon modes S9, S10, S15, S16 S17, S18, S19 and S20 show normal
temperature dependence. (ii) FWHM of the modes S9, S16, S17, S19, S20 show normal
temperature dependence. FWHM of S10 modes show normal behaviour only down to ~250 K, and
below 200 K, it remains nearly temperature-independent down to the lowest recorded temperature
(4 K). FWHM of S15 mode increases slightly with decreasing temperature till ~ 200 K; on further
cooling, it starts to decrease. FWHM of S18 mode increases with decreasing temperature from 330
K to lowest recorded temperature, 4 K. The temperature dependence of the mode frequency ( )
and FWHM of S9, S10, S15, S16 S17, S18, S19 and S20 phonon modes for the bilayer is shown
in Fig. S7. Following observations can be made: (i) Frequency of all phonon modes S9, S10, S15,
S16 S17, S18, S19 and S20 except S9 show normal temperature dependence. (ii) FWHM of the
modes S9, S10, S15, S16, S17, S19 show normal temperature dependence. FWHM of S18 mode
increases with decreasing temperature from 330 K to lowest recorded temperature, 4 K, FWHM
of S20 mode first decreases with decreasing temperature down to ~100 K, on further cooling, it
starts to increase. The solid red lines in Fig. S5, Fig.S6 and Fig. S7 are the fitted curves via three
phonon anharmonic model as described in the main text, and the fitting is in good agreement with
the experimental data above ~100 K, and the solid green lines are guide to the eye.
39
FIGURE S7: Temperature dependence of the phonon mode frequencies and FWHM of the modes
(a) S9, S10, S15, and S16; (b) S17, S18, S19 and S20 for bilayer MoSe2. Solid red lines are the
fitted curves via three phonon anharmonic model as described in the main text and solid green
lines are guide to the eye.
4. Temperature-dependent Raman intensity of the phonon modes
In most of the previous literature, we note that the Raman intensity of the phonons and its
dependence on the temperature has not been touched upon. While, in these 2D materials, the
temperature-dependent Raman intensity of the phonon modes may provide rich information about
the electronic and optical properties of the materials as well as tuning of these properties as a
function of temperature. Very few reports for these 2D materials are available to the best of our
knowledge, wherein Raman intensity of the phonon modes and its dependence on the temperature
has been studied in detailed [2-3]. However, no such reports are available for the case of MoSe2.
This section focuses on the Raman intensity of the phonon modes and its temperature dependence
for mono and bilayer. From Fig. S4 (a) and S4 (b), one could see that S6 (1gE ) mode is prominent
at high temperature in monolayer as well as bilayer, while S8 ( 1
2gE ) mode is found to be intense at
40
high temperature only for monolayer and it becomes intense at low temperature for the bilayer.
Temperature-dependent intensity evaluation of the S12 ( 2
2uA ) mode is not very clear for the case of
the monolayer, maybe because of the very weak signal, while for the bilayer, one could see that it
become intense at low temperature compared to that of high temperature. To quantitatively
understand the temperature-dependent Raman intensity of the phonon modes, we extracted the
intensity of the individual phonons by using the Lorentzian function. Figure S8 (a) and S8 (b)
show the integrated intensity of the few prominent modes such as S5 ( LA ), S6 (1gE ), S7 (
1gA ), S8
( 1
2gE ) and S12 ( 2
2uA ) as a function of temperature for monolayer and bilayer, respectively. For the
monolayer, we observed that the intensity of the S5 ( LA ), S6 (1gE ), S8 ( 1
2gE ) and S12 ( 2
2uA ) modes
decrease with decreasing temperature, while it remains temperature independent for the S7 (1gA )
mode. For bilayer, the intensity of the S5 ( LA ), S6 (1gE ), S8 ( 1
2gE ) and S12 ( 2
2uA ) modes increases,
while it decreases for the S6 (1gE ), S7 (
1gA ) modes, with decreasing temperature. Figure S8 (c)
shows the intensity ratio of the modes as a function of temperature for monolayer and bilayer. We
observed that the intensity ratio of the 1gA mode with respect to 1
2gE mode increases (decreases)
with decreasing temperature for monolayer (bilayer), while the intensity of the 2
2uA mode with
respect to 1gE mode increases for both layers. The intensity ratio of the 2LA mode with respect to
LA increases (decreases) with the decrease in temperature for the monolayer (bilayer), while an
increase is observed in the intensity ratio of the 3LA and 4LA modes with respect to LA mode with
decreasing temperature in both these systems. At low temperature, 3LA and 4LA modes are ~ 4-5
(~ 2) times stronger compared to that of LA mode for the monolayer (bilayer). 2
2uA mode is ~ 4-5
times stronger than 1gE in both monolayer and bilayer at low temperature. On the other hand,
1gA
41
mode is ~14-15 times stronger than that of 1
2gE mode at low temperature for the monolayer, while
it is ~12-13 times stronger than that of 1
2gE mode at high temperature for the bilayer.
Now we focus on understanding the temperature-dependent variations in the intensity of the
phonon modes. In these kinds of 2D materials, the variations in the intensity of the phonon modes
as a function of temperature preliminary may be understood considering the tuning of resonance
conditions with temperature variations. The tuning of the resonance conditions as a function of
incident photon energies and its impact on the intensity of the phonon modes in these TMDCs
materials, including MoSe2, has been studied in several earlier studies [4-5]. Phonon modes1gA ,
1
2gE and LA and their overtone become prominent when excitation energies resonate with the A and
B excitonic energy band. In contrast, the intensity of the 1gE and 2
2uA modes is enhanced when laser
energy resonates with the C exciton energy. In the quantum mechanical picture, the Raman
scattering intensity of phonon mode may be given as [3, 6]
2
1
[ ( ) ( )][ ( ) ( )]L s
IE T E i T E T E i T
(1)
where ( )E T and ( )T are the temperature-dependent transition energies and damping constants of
excitonic bands, respectively, while LE (
SE ) is the energy of the incident (scattered) photons, which
is fixed in our case. The energy of the excitons are strongly dependent on temperature, and it could
be tuned as a function of temperature. Therefore, in the present case as well, the resonance
conditions may be changed by changing the temperature of the sample. The energy of the excitonic
bands increases with descreasing temperature [7]. As the temperature increases, the energy of C
exciton move away from the incident laser photon energy, while at the same time the energies of
A or B excitons approachs the incident laser photon energy.
42
Therefore, the intensity of the modes1gA , 1
2gE and LA as well as its overtone, should be expected
to increase with decreasing temperature, while the opposite nature is anticipated for the 1gE and
2
2uA modes. The temperature-dependent intensity of 1gE mode for both layers is in good agreement
with the above statement, but 2
2uA mode only shows good agreement for monolayer, while it
violates for bilayer, see Fig. S8 (a) and S8 (b). The intensity of the 1gA , 1
2gE and LA also shows
different temperature dependence for different thickness. For example 1gA mode is intense in
bilayer compared to monolayer, while opposite nature is observed for the 1
2gE mode (see Fig. 3 in
main text). Therefore, the observed different temperature-dependent intensity for different modes
and different layer thickness could not be explained, considering only the tuning of the resonance
condition. It is reported that for these 2D materials, the wave function of A and B exciton are
strongly confined to the single individual layer, with only a slight overlap to the adjoining layers
[8]. On the other hand, the wave function of the C exciton extended over the entire thickness in
the layer system. Moreover, the energy of A and C excitons is observed to depend strongly on the
number of layers, while the energy of B exciton shows very weak dependence. The spatial
confinement of the wave function of the excitons and layers dependence energies of the excitons
possesses different resonance effect for different phonon modes. This may give the different
temperature and thickness dependence of the intensity of the phonon modes. Further, TEC and
TEC mismatch between MoSe2 and substrate also significantly impact the intensity of the phonon
modes. Because these effects are strongly dependent on the layer thickness and temperature, and
these may be other possible reasons for different temperature and different thickness-dependent
intensity of the observed phonons modes in MoSe2 system studied here.
43
FIGURE S8: (Color online) (a) and (b) Temperature dependence of the integrated intensity of S5
( LA ), S6 (1gE ), S7 (
1gA ), S8 ( 1
2gE ) and S12 ( 2
2uA ) phonon modes for monolayer and bilayer,
respectively. (c) Temperature dependence of the integrated intensity ratio of the phonon modes.
References:
[1] N. Scheuschner et al., Phys. Rev. B 91, 235409 (2015).
[2] D. Kumar et al., Nanotechnology. 32, 285705 (2021).
[3] H. Zobeiri et al., Nanoscale 12, 6064 (2020).
[4] K. Kim et al., ACS Nano 10, 8113 (2016).
44
[5] P. Soubelet et al., Phys. Rev. B 93, 155407 (2016).
[6] Light Scattering in Solid II, edited by M. Cardona and G. Guntherodt, Springer Verlag
Berlin (1982).
[7] H.G. Park et al., Sci. Rep. 8, 3173 (2018).
[8] G.Y. Jia et al., Mater. Chem. C 4, 8822 (2016).